Nanopore sequencing and full genome de novo assembly of human cytomegalovirus TB40/E reveals clonal diversity and structural variations

Timokratis Karamitros,Bonnie Van Wilgenburg,Gkikas Magiorkinis,Mark Wills,Paul Klenerman

doi:10.1186/s12864-018-4949-6

Abstract

BackgroundHuman cytomegalovirus (HCMV) has a double-stranded DNA genome of approximately 235 Kbp that is structurally complex including extended GC-rich repeated regions. Genomic recombination events are frequent in HCMV cultures but have also been observed in vivo. Thus, the assembly of HCMV whole genomes from technologies producing shorter than 500 bp sequences is technically challenging. Here we improved the reconstruction of HCMV full genomes by means of a hybrid, de novo genome-assembly bioinformatics pipeline upon data generated from the recently released MinION MkI B sequencer from Oxford Nanopore Technologies.ResultsThe MinION run of the HCMV (strain TB40/E) library resulted in ~ 47,000 reads from a single R9 flowcell and in ~ 100× average read depth across the virus genome. We developed a novel, self-correcting bioinformatics algorithm to assemble the pooled HCMV genomes in three stages. In the first stage of the bioinformatics algorithm, long contigs (N50 = 21,892) of lower accuracy were reconstructed. In the second stage, short contigs (N50 = 5686) of higher accuracy were assembled, while in the final stage the high quality contigs served as template for the correction of the longer contigs resulting in a high-accuracy, full genome assembly (N50 = 41,056). We were able to reconstruct a single representative haplotype without employing any scaffolding steps. The majority (98.8%) of the genomic features from the reference strain were accurately annotated on this full genome construct. Our method also allowed the detection of multiple alternative sub-genomic fragments and non-canonical structures suggesting rearrangement events between the unique (UL /US) and the repeated (T/IRL/S) genomic regions.ConclusionsThird generation high-throughput sequencing technologies can accurately reconstruct full-length HCMV genomes including their low-complexity and highly repetitive regions. Full-length HCMV genomes could prove crucial in understanding the genetic determinants and viral evolution underpinning drug resistance, virulence and pathogenesis.

Highlights

Human cytomegalovirus (HCMV) has a double-stranded DNA genome of approximately 235 Kbp that is structurally complex including extended GC-rich repeated regions
The 235 Kbp-long double-stranded viral genome is partitioned in two major segments, the Unique Long (UL) and the Unique Short (US) both flanked by terminal -T- and internal -I- repeated sequences (TLR/ILR, Terminal and Internal Short Repeated sequences (TSR/ISR)) [2]
Just like HHV-1 [37], HCMV’s class-E genome is known to create isomers [2] and here, we prove for the first time that MinION captures such structures in a much simpler way compared to complex molecular techniques

Summary

Introduction

Human cytomegalovirus (HCMV) has a double-stranded DNA genome of approximately 235 Kbp that is structurally complex including extended GC-rich repeated regions. The 235 Kbp-long double-stranded viral genome is partitioned in two major segments, the Unique Long (UL) and the Unique Short (US) both flanked by terminal -T- and internal -I- repeated sequences (TLR/ILR, TSR/ISR) [2]. These two genomic segments may invert with respect to each other resulting in four genomic isomers, which can be present in equal concentrations [2] suggesting that HCMV is characterized by extended structural genomic plasticity. Clones “Lisa” and “Bart” were originally plaque-picked from TB40/E cultures to isolate clones that could or could not evade Natural Killer (NK) cells function [9]

Methods

Results

Discussion

Conclusion